Considering critical building materials for embodied carbon emissions in buildings: A machine learning-based prediction model and tool

Construction activities discharge considerable carbon emissions, causing serious environmental problems and gaining increasing attention. For the large-scale construction area, high emission intensity, and significant carbon reduction potential, embodied carbon emissions of buildings worth special studying. However, previous studies are usually post-evaluation and ignore the influences of project, construction and field. This paper focuses on critical building materials and adopts machine learning methods to realize carbon prediction at design stage. The activity data, including critical building materials, water, and energy consumption are analyzed and 30 influencing factors at the project, construction, and management levels are identified. Three algorithms (artificial neural network, support vector regression and extreme gradient boosting) are used to develop machine learning models. The proposed methodology is applied to 70 projects in the Yangtze River Delta region of China. Results show that the established model achieved high interpretability (R2>0.7) and small average error (5.33%), well proving its feasibility. Furthermore, an automated tool is developed to assist practitioners to predict the critical materials consumption and embodied carbon emissions conveniently. The operable model and practical tool can efficiently predict critical material consumption and embodied carbon emissions at design stage, supporting effective adjustments and improvement to reduce carbon in construction.